I distinctly remember the first time I tried to write a job post with AI. I was feeling nervous for the same reason most recruiters do right now, wondering, “is this going to replace me?” Then I actually tried to write a job post and my worries were erased. I laughed out loud, actually. The post was so bad. It was full of buzzwords, bias, and bullshit. Checked almost every single box of my scary job post criteria.

Job posts are some of the most important recruitment marketing copy anyone could ever write - statistically and emotionally. Look at the source of hire data. Regardless of your source, whether it’s public resources like BLS.gov or private resources like LinkedIn, job boards always show up in the top 5 sources of hire. Typically, top 3. The majority of people who find jobs in this country still find them by reading a job post and applying. 

Emotionally, what part of the job search *isn't* loaded with emotions? Do you remember finding a job post for a role you were really excited about at some point in your life? The giddy joy you feel finding it in the heap of roles that don’t work for you. That moment is exciting. Those words could change your life - and I'm not saying that for dramatic affect. When you change where you go every day, who you talk to, and even the computer you use to work with one choice? That's life-altering.

Why AI Job Posts (Still) SUCK 

Still, job posts are as awful as they were before AI. For the last 20+ years, job boards have been full of average to awful, generic job posts with exaggerated requirements. Enter AI and a lot of relief from people who thought that was the end of their job post writing days. At no point did these folks stop to think about the training data used to write these new job posts: the same junk everyone else has been posting on job boards for the last 10 years. Of course it doesn't generate better job posts. The data set is trash.

In the past few years, I’ve seen more evolved attempts at using AI to write job posts like job post graders, using transcripts to generate the text, and job posts written by AI assistants. While these tools can make some awful job posts marginally better by editing things out that would have been caught by any human editor, they still miss one big mark: they don't stand out. They all sound exactly the same. Same buzzwords. Same phrases. Same BS. These job posts are creating a sea of sameness where your posts say a lot without saying anything at all.

But really, sounding the same isn't even the critical issue here. It's the fact that most of these AI generated job posts are too generic to generate results. These job posts are AI illusions of a job, not reality - and that’s an expensive problem when these illusions are used to inform AI agents that screen hires. If you can't tell the agent what you need clearly in the job description, it'll reject the people you want. That’s an expensive mistake just to save yourself 30 minutes on a job post.

Four Ways To Improve A Job Post With AI 

That doesn't mean you can't write a job post with AI or it's ruined forever. It means we need to use AI as an editor, not a creator in the job post creation process. Here are a few guidelines that will help you use AI to generate something more useful than the generic “write me a job post” prompt. 

1. Never prompt for a first draft by saying “write me a [job title] job post.” All job titles are made up and there’s not always consistency in roles with the same title with the exception of jobs like doctor or lawyer. If you’re going to prompt for a first draft, be sure to include every day activities and mandatory requirements instead of the job title. 
2. Voice to text your first draft and don’t hesitate to be verbose about what you’re really looking for. Recruiters are really good at explaining a role when I talk to them, but not when I ask them to write it down. Use that to your advantage. Talk through the first drafts, then give your LLM the template and tell it to write a first draft. Use prompts like "make this shorter" or "create clarity" to take your long version and turn it into something more usable.
3. Never copy and paste a final draft from an open LLM window into the posting tool. Post into a Word or Google doc first. Read it again. Look for placeholders and the “next step” prompts ChatGPT always adds to the end. A human should always be the final review to ensure AI hasn’t edited the personality and details out of your post. Remember, you're only one f away from being the shit leader.
4. Get better at asking questions. The more questions and clarity you have about a role after the hiring manager intake, the better. No matter what prompt anyone gives you, you can’t generate anything useful with AI if you can’t define what the high quality output is. 

If you’re looking for a step by step guide on how to write a better job post, download my free guide or take the on-demand course where I’ll walk you through step by step how to create better job posts and define roles.